0
Research Papers

# The Influence of the Joint Wind-Wave Environment on Offshore Wind Turbine Support Structure Loads

[+] Author and Article Information
Puneet Agarwal, Lance Manuel

Department of Civil, Architectural, and Environmental Engineering, University of Texas, Austin, TX 78712

As an alternate to the Rayleigh distribution, a Weibull distribution for the mean wind speed could also be employed. This is a more general distribution and the only change would be in Eq. 2 where the exponent, 2, could be replaced by a parameter $k(θ)$.

J. Sol. Energy Eng 130(3), 031010 (Jul 02, 2008) (11 pages) doi:10.1115/1.2931500 History: Received January 18, 2007; Revised September 01, 2007; Published July 02, 2008

## Abstract

Our objective here is to establish long-term loads for offshore wind turbines using a probabilistic approach. This can enable one to estimate design loads for a prescribed level of return period, generally on the order of $20–50years$ for offshore wind turbines. In a probabilistic approach, one first needs to establish “short-term” distributions of the load random variable(s) conditional on the environment; this is achieved either by using simulation or field measurements. In the present study, we use field data from the Blyth offshore wind farm in the United Kingdom, where a $2MW$ wind turbine was instrumented, and environment and load data were recorded. The characteristics of the environment and, hence, that of the turbine response at the site are strikingly different for wind regimes associated with different wind directions. Here, we study the influence of such contrasting environmental (wind) regimes and associated waves on long-term design loads. The field data, available as summary statistics, are limited in the sense that not all combinations of environmental conditions likely to be experienced by the turbine over its service life are represented in the measurements. Using the available data, we show how distributions for random variables describing the environment (i.e., wind and waves) and the turbine load of interest (i.e., the mudline bending moment) can be established. By integrating load distributions, conditional on the environment with the relative likelihood of different environmental conditions, long-term (extreme/ultimate) loads associated with specified return periods can be derived. This is demonstrated here by carefully separating out the data in different wind direction sectors that reflect contrasting wind (and accompanying wave) characteristics in the ocean environment. Since the field data are limited, the derived long-term design loads have inherent uncertainty associated with them; we investigate this uncertainty in such derived loads using bootstrap techniques.

<>

## Figures

Figure 1

(a) Location of the turbines and an onshore meteorological mast at the Blyth site; (b) layout of the turbine instrumentation (from Camp (5))

Figure 2

(a) Histogram showing number of occurrences or data sets for different mean wind directions while the turbine is operating; (b) histogram showing division of data according to both mean wind direction and mean wind speed

Figure 3

Scatter diagram showing mean wind speed versus significant wave height

Figure 4

(a) Storm profile and (b) wind-wave scatter diagram for winds from the sea. The duration of the storm shown is from Dec. 7 to Dec. 10, 2002.

Figure 5

(a) Storm profile and (b) wind-wave scatter diagram for winds from the shore. The duration of the storm shown is from Jan. 22 to Jan. 24, 2003.

Figure 6

Wind-wave scatter diagrams for each of the 30deg wind-direction, θ, sectors

Figure 7

Sample averages and sample 90% confidence intervals (5- and 95-percentile levels) of 10min mean wind speed for different mean wind directions

Figure 8

Weibull distribution parameters for significant wave height conditional on mean wind speed and mean wind direction—(a) shape parameter, k; (b) scale parameter, λ. Estimates are for the 90–120deg wind-direction sector representative of winds from the sea and the 240–270deg wind-direction sector representative of winds from the shore.

Figure 9

(a) Ratio of the SRSS of the maximum mudline bending moments in two orthogonal directions to the maximum vector-resultant mudline bending moment; (b) variation of the mean direction of the vector-resultant mudline bending moment with the mean wind direction

Figure 10

Estimates of the Gumbel distribution parameters, u and β, for maximum mudline bending moment in (a) the 90–120deg wind-direction sector and (b) the 240–270deg wind-direction sector

Figure 11

Comparison of probability of load exceedance curves for winds from the sea, winds from the shore, and winds from all directions

Figure 12

Design loads (mudline bending moments) for (a) 1year and (b) 20year return periods for different wind-direction sectors. Also shown are 90% confidence intervals (5- and 95-percentile design load levels) along with mean values for the “with-uncertainty” case based on bootstrap methods applied to the limited data in each sector.

Figure 13

Comparison of 90% confidence intervals and mean value estimates of the 20year design load based on different methods of estimation of parameters of the short-term load distribution: (a) regression over the largest above-median loads data; (b) regression over all the loads data; (c) collocation using the 50th and 90th load percentiles; and (d) the method of moments

Figure 14

Comparison of 20-year design loads when the environmental random variable vector, X, consists of (a) mean wind speed (V) and significant wave height (Hs); and (b) mean wind speed (V) and turbulence standard deviation (σV)

## Discussions

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related Proceedings Articles
Related eBook Content
Topic Collections